EEVblog Electronics Community Forum
Electronics => Microcontrollers => Topic started by: diyaudio on June 27, 2015, 10:36:55 pm
-
Can anyone recommend a good 1.1V @ 1.5A LDO to power a processor core, I cannot seem to find many LDO solutions, Google search results are 90% buck based controllers and my aim is to keep the BOM and complexity down.
So far I only found this, and its in those pesky QFN (not a big issue, just a dog to hand solder)
http://www.ti.com/lit/ds/symlink/tps74801.pdf (http://www.ti.com/lit/ds/symlink/tps74801.pdf)
Can anyone recommend alternatives ?
-
At 1.1V and 1.5A your going to drop voltage pretty fast just getting the power into the IC
Maybe a 1.2V reg would be better? Whats the max input voltage of your processor?
Quick digikey search found a 1.2V 1.5A Vreg pretty easy. Min input is 2.2V
http://www.digikey.com/product-detail/en/AP7363-12D-13/AP7363-12D-13CT-ND/3829380 (http://www.digikey.com/product-detail/en/AP7363-12D-13/AP7363-12D-13CT-ND/3829380)
-
What about something like LT3081 or LT3083 in TO220?
With best regards
Andreas
-
At 1.1V and 1.5A your going to drop voltage pretty fast just getting the power into the IC
Maybe a 1.2V reg would be better? Whats the max input voltage of your processor?
Quick digikey search found a 1.2V 1.5A Vreg pretty easy. Min input is 2.2V
http://www.digikey.com/product-detail/en/AP7363-12D-13/AP7363-12D-13CT-ND/3829380 (http://www.digikey.com/product-detail/en/AP7363-12D-13/AP7363-12D-13CT-ND/3829380)
Maximum input is 5V. The normal processor core supply voltage is 1.1V, max is 1.15V.
-
There's a reason why you only find switching regulators at that power level.
You will need to dissipate nearly 6W of heat, to run a device that will use 1.65W at most. That is a horrible waste of power.
The AP7363 mentioned above has a best-case thermal resistance of 52.8C/W junction to ambient, meaning that it would go over 300C above ambient if it were made to dissipate 6W. The TPS74801 is not much better, at 35.6C/W.
-
LDO for a processor core? I bet it doesn't drive any high precision analog blocks, so forget using LDOs. DC/DC converters are used for that purpose :rant:
Check TI or Linear DC/DCs, proven to be good :-+
-
I like ST1S10 for applications like that. Small inductor and easy application schematic. And it's cheap. Low voltage buck regulators are nowadays really easy to implement.
Sent from my HTC One M8s using Tapatalk
-
Even the 20-pin TPS74801 QFN variant will probably overheat with the load proposed.
E.g.
Vin=5V, Vout=1.1V, Iout=1A (assuming "average") -> (5-1.1)*1 = 3.9W
3.9W * 35.6C/W + 30C ambient = 168C
Absolute max is 150C, probably want to keep it under 100C if you want to operate for a long period of time.
It would probably work if you could get a TO220 LDO regulator like a LT3080 and put it on a small heatsink.
In that case I think a LDO buck is much simpler and cheaper. Just find one that can down to 0.8V (edit: as often others get stuck at 1.2V or so) and re-cycle the 1.2V application schematic
-
Even the 20-pin TPS74801 QFN variant will probably overheat with the load proposed.
E.g.
Vin=5V, Vout=1.1V, Iout=1A (assuming "average") -> (5-1.1)*1 = 3.9W
3.9W * 35.6C/W + 30C ambient = 168C
Absolute max is 150C, probably want to keep it under 100C if you want to operate for a long period of time.
It would probably work if you could get a TO220 LDO regulator like a LT3080 and put it on a small heatsink.
In that case I think a LDO is much simpler and cheaper. Just find one that can down to 0.8V and re-cycle the 1.2V application schematic
You guys are 100% percent correct, my calculations were based on dynamic power, however I never considered what happens if I ran the core continuously with some lengthy floating point computations |O
I I will settle for a buck then, that ST1S10 looks nice and simple (thanks!), however now im spoiled for choice :)
maybe a dual (with integrated FETS) 1.1v @2A, and 3.3V @ 1A will be better, I can then eliminate my 3.3V LDO and slip one solution..
-
IMO considering LDO that will dissipate anything more than 0.5W or so is pointless unless it's a lab power supply or you are powering very sensitive analog stuff.
Sent from my HTC One M8s using Tapatalk
-
Is the 1v1 for a Sharc? If so, I would highly recommend the regulator circuit from the dev board. We've done many Sharc designs, and always come back to the ref design, having lost many other attempted cheaper/smaller circuits to the magic smoke.
-
Is the 1v1 for a Sharc? If so, I would highly recommend the regulator circuit from the dev board. We've done many Sharc designs, and always come back to the ref design, having lost many other attempted cheaper/smaller circuits to the magic smoke.
Yep, SHARC 21489 there are a few options, magic smoke?, its rather expensive smoke :)
-
This is the default circuit I've been using to power two 21489 devices in various products for over 3 years with zero failures. It's probably more involved than you want, but is guaranteed to work.
-
If you really want linear, a good IC would be LT3081 (http://www.digikey.com/product-search/en/integrated-circuits-ics/pmic-voltage-regulators-linear-ldo/2556290?k=lt3081) but of course, it's expensive ( ~ 6$ for one). LT3083 would be good for 3A but it's close to 10$.
-
If you really want linear, a good IC would be LT3081 (http://www.digikey.com/product-search/en/integrated-circuits-ics/pmic-voltage-regulators-linear-ldo/2556290?k=lt3081) but of course, it's expensive ( ~ 6$ for one). LT3083 would be good for 3A but it's close to 10$.
Linear is always the most expensive option. I've successfully used MPS (https://www.monolithicpower.com/). Now I mainly rely on TI products. The designs are automotive, so passes all the tests and must last 15 years.
-
This is the default circuit I've been using to power two 21489 devices in various products for over 3 years with zero failures. It's probably more involved than you want, but is guaranteed to work.
Thanks!